Advice on BGA device pad design , layout and routing...

peter wrote:

All,

 

I've been using through-hole and lately SMD devices extensively on one

and two layer boards, but I'm about to make my first foray into (what

I consider to be ) big BGA components and thats pushing me to 4 layers

(hopefully not higher) boards and more complexity.

 

So I'm looking for experienced viewpoints... since the excercies has

brought several questions up when I've been thinking about this...

 

...

 

- I think I can get away with a 4 layer board, 4 layers should be

enough for me to route all the pins I need, BUT... will the board

houses allow me to put a bunch of signals on the two internal layers that

they seem

to expect are power supply and ground planes..., is this just

terminology/convention or is there a real limitation here.

You can route anything you like on the inner layers.  Don't define them as

power planes in Eagle.  Instead, just place GND and VCC polygons where you

need them.  That way you can place other traces in the inner layers, and

Eagle will fill the remaining space with your power planes.

- I really don't want to have to think about a 6 layer design since

that means a more expensive Eagle license and a more expensive board

manufacturing cost.. this is all hobby based stuff..

I have not done any BGA layout myself, but have been involved with the

assembly of some prototype boards with BGA packages at an assembly house

that was just getting started with BGA assembly.  No end of trouble, despite

all the proper equipment they had (including X-ray inspection).  I don't

think BGA assembly is suited for hobbyist DIY assembly.  Are you sure you

can't this CPU in some less painful package?  Or maybe you can buy an

adapter?

--

Bert Menkveld

Bert Menkveld schrieb:

all the proper equipment they had (including X-ray inspection).  I don't

think BGA assembly is suited for hobbyist DIY assembly.  Are you sure you

Do you know http://www.lrr.in.tum.de/~acher/bga/index.html? It is in

German but maybe you get an idea of how to solder a BGA with an

electric iron and some other simple stuff most of us have at home.

--

Matthias Weißer

matthias@matwei.de

http://www.matwei.de

Matthias Weisser wrote:

Bert Menkveld schrieb:

all the proper equipment they had (including X-ray inspection).  I

don't think BGA assembly is suited for hobbyist DIY assembly.  Are

you sure you

 

Do you know http://www.lrr.in.tum.de/~acher/bga/index.html? It is in

German but maybe you get an idea of how to solder a BGA with an

electric iron and some other simple stuff most of us have at home.

I don't understand much German, but I'll take your word for it that it is

possible to solder your own BGA's.  My hat off to you and all those other

amazing hobbyists out there who manage what I thought pretty near

impossible.  Guess I've been a professional too long to realize what a

determined amateur can do.

--

Bert Menkveld

Matthias Weisser wrote:

Bert Menkveld schrieb:

all the proper equipment they had (including X-ray inspection).  I

don't think BGA assembly is suited for hobbyist DIY assembly.  Are you

sure you

 

Do you know http://www.lrr.in.tum.de/~acher/bga/index.html? It is in

German but maybe you get an idea of how to solder a BGA with an electric

iron and some other simple stuff most of us have at home.

 

If you type the link into the Google search window and search for it,

you will get only one hit - Then click on the 'Translate this page' link

and you will get the webpage (somewhat) translated into English. It is

good enough to get a pretty good idea of what is being said.

Jack

Matthias Weisser wrote:

Bert Menkveld schrieb:

all the proper equipment they had (including X-ray inspection).  I

don't think BGA assembly is suited for hobbyist DIY assembly.  Are you

sure you

 

Do you know http://www.lrr.in.tum.de/~acher/bga/index.html? It is in

German but maybe you get an idea of how to solder a BGA with an electric

iron and some other simple stuff most of us have at home.

 

Google has a reasonable attempt at translating the site, but there is

definitely some guess work to be done after the translation

The referenced Motorola App Note if very good, I'll be using that as a

reference.

I've not tried this yet, but my experience with using solder paste on my

boards with surface mount parts has been exceptionally good, dabbing a

little paste from a modellers glue dispenser (syringe with small needle)

on the pads, placing the component with tweezers on the pasted pad

populating the SMD parts tyhis way then putting the whole aseembly

(carefully) into a toaster oven (with a fan to circulate the air to

reduce hotspots) , cranking the heat to 240 Centigrade and back down.

The discrete SMD's look very professionally soldered, the whole board

looks clean and finished with no extra work, and I've never had a

problem with them, some of the higher pin density SOIC type devices may

bridge between pins if I applied too much paste, but its easily fixed by

a little solder wick and application of a soldering iron to the affected

pins.

My reading up on this says the same principles apply with BGA devices,

with the exception that bridges can't be fixed... its a one chance deal.

SO... the recommendation appears to be apply the very small amount of

paste(or even just flux) by using an appropriate stencil (perhaps for

just the BGA, but why not the whole board kits can be had fairly cheap

now, this gets you just the the right amount of paste onto the pads

through the stencil holes...careful placement of the BGA and then a

better controlled heating /cooling cycle should have the desired effect.

Once I get my board routed I'll be able to tell you if it works... I do

envisage a steep learning curve, I have three processors to work with,

at $20 each I can afford to learn from a couple of mistakes.

Peter

Matthias Weisser wrote:

Bert Menkveld schrieb:

all the proper equipment they had (including X-ray inspection).  I

don't think BGA assembly is suited for hobbyist DIY assembly.  Are you

sure you

 

Do you know http://www.lrr.in.tum.de/~acher/bga/index.html? It is in

German but maybe you get an idea of how to solder a BGA with an electric

iron and some other simple stuff most of us have at home.

 

Google has a reasonable attempt at translating the site, but there is

definitely some guess work to be done after the translation

The referenced Motorola App Note if very good, I'll be using that as a

reference.

I've not tried this yet, but my experience with using solder paste on my

boards with surface mount parts has been exceptionally good, dabbing a

little paste from a modellers glue dispenser (syringe with small needle)

on the pads, placing the component with tweezers on the pasted pad

populating the SMD parts tyhis way then putting the whole aseembly

(carefully) into a toaster oven (with a fan to circulate the air to

reduce hotspots) , cranking the heat to 240 Centigrade and back down.

The discrete SMD's look very professionally soldered, the whole board

looks clean and finished with no extra work, and I've never had a

problem with them, some of the higher pin density SOIC type devices may

bridge between pins if I applied too much paste, but its easily fixed by

a little solder wick and application of a soldering iron to the affected

pins.

My reading up on this says the same principles apply with BGA devices,

with the exception that bridges can't be fixed... its a one chance deal.

SO... the recommendation appears to be apply the very small amount of

paste(or even just flux) by using an appropriate stencil (perhaps for

just the BGA, but why not the whole board kits can be had fairly cheap

now, this gets you just the the right amount of paste onto the pads

through the stencil holes...careful placement of the BGA and then a

better controlled heating /cooling cycle should have the desired effect.

Once I get my board routed I'll be able to tell you if it works... I do

envisage a steep learning curve, I have three processors to work with,

at $20 each I can afford to learn from a couple of mistakes.

Peter

Bert Menkveld wrote:

peter wrote:

All,

 

I've been using through-hole and lately SMD devices extensively on one

and two layer boards, but I'm about to make my first foray into (what

I consider to be ) big BGA components and thats pushing me to 4 layers

(hopefully not higher) boards and more complexity.

 

So I'm looking for experienced viewpoints... since the excercies has

brought several questions up when I've been thinking about this...

 

...

 

- I think I can get away with a 4 layer board, 4 layers should be

enough for me to route all the pins I need, BUT... will the board

houses allow me to put a bunch of signals on the two internal layers that

they seem

to expect are power supply and ground planes..., is this just

terminology/convention or is there a real limitation here.

 

You can route anything you like on the inner layers.  Don't define them as

power planes in Eagle.  Instead, just place GND and VCC polygons where you

need them.  That way you can place other traces in the inner layers, and

Eagle will fill the remaining space with your power planes.

Thanks, that what I wanted to hear

- I really don't want to have to think about a 6 layer design since

that means a more expensive Eagle license and a more expensive board

manufacturing cost.. this is all hobby based stuff..

 

I have not done any BGA layout myself, but have been involved with the

assembly of some prototype boards with BGA packages at an assembly house

that was just getting started with BGA assembly.  No end of trouble, despite

all the proper equipment they had (including X-ray inspection).  I don't

think BGA assembly is suited for hobbyist DIY assembly.  Are you sure you

can't this CPU in some less painful package?  Or maybe you can buy an

adapter?

 

It seems that there will be some learning to get to a reliable process

for this, and there is no satisfactory home grown test for BGA soldering

results other than a full functional test...that I've come across so far

anyway.

Unless you know how to make a cheap (and safe) Xray machine 

--

Bert Menkveld

 

 

Thanks,

Peter

peter wrote:

Matthias Weisser wrote:

my experience with using solder paste on my

boards with surface mount parts has been exceptionally good, dabbing a

little paste from a modellers glue dispenser (syringe with small needle)

on the pads, placing the component with tweezers on the pasted pad

populating the SMD parts tyhis way then putting the whole aseembly

(carefully) into a toaster oven (with a fan to circulate the air to

reduce hotspots) , cranking the heat to 240 Centigrade and back down.

It you ever need, there exist brass sheet with photo resist so you can

etch your own pasta stencil.

IIRC http://www1.business.conrad.de have them.

Or maybe your board house can make one.

The discrete SMD's look very professionally soldered, the whole board

looks clean and finished with no extra work, and I've never had a

problem with them, some of the higher pin density SOIC type devices may

bridge between pins if I applied too much paste, but its easily fixed by

a little solder wick and application of a soldering iron to the affected

pins.

I have gotten up to speed on prototype series using a good 0,25 mm lead

free solder on my normal 0603 and SOIC.  The 0,5mm spacing pins i solder

over all, then wick it back.

For the few SMT resistor nets with pads underneath i have used solder

paste and a cheap hot air solder gun, also from Conrad.

For one-off BGA i think my personal take would be to cut a big hole in

the board and solder thin enamelled wires for the signals and some mesh

for GND and then decoupling caps "on top" of that... mess...

I have hand routed a pretty dense SOIC prototype using thin enameled

wires only.  In this case yoi really need good solder, wire and solder

iron, and plan the work moving soldering from one side to the other in

sequence...